Ultrasensitive Detection of PD-L1-Expressing Extracellular Vesicles via Reflection-Mediated Ellipsometry: From Nanoscale Optical Modeling to Immunotherapy Response Prediction.

Programmed death-ligand 1 (PD-L1) expression in extracellular vesicles (EVs) has emerged as a promising biomarker for predicting responses to immune checkpoint inhibitors (ICIs), but current detection methods lack the sensitivity needed for clinical translation. Here, a reflection-mode solution-immersed silicon (R-SIS) ellipsometry sensor is adapted, originally developed for thin-film measurements, for the ultra-sensitive detection of PD-L1-expressing EVs. Computational and experimental validation demonstrated that the selective use of p-polarized light, combined with a fixed polarizer configuration, enables highly sensitive detection of nanoscale surface changes. Applied to cell-line-derived EVs, the sensor achieved a limit of detection of 3.8 × 10 particles mL-exceptionally high sensitivity for label-free EV detection-with also high reproducibility and specificity. In clinical samples from patients receiving ICI therapy, PD-L1⁺ EV levels measured by the R-SIS sensor distinguished responders from non-responders and correlated with progression-free survival, outperforming conventional ELISA and tissue-based PD-L1 staining. Additionally, it is demonstrated that direct detection of intact EVs yields stronger signals than lysate-based detection due to amplified optical responses from vesicle-scale binding. The platform's versatility is further demonstrated by detecting alternative tumor-derived EV biomarkers and employing diverse capture ligands. These findings establish the R-SIS sensor as a clinically actionable platform for high-sensitivity, EV-based liquid biopsy.